fmnist.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "fmnist.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/semla/notebooks/blob/master/fmnist.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VtEcg8p3gR3_",
        "colab_type": "text"
      },
      "source": [
        "# Install dependencies & import\n",
        "Install Keras and Tensorflow using pip."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rjxoiYwLfV3A",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "a55b74aa-a5a1-4633-aac6-4289fa46eb93"
      },
      "source": [
        "# !pip install -U keras tensorflow>=1.8.0 matplotlib numpy\n",
        "!pip install -U keras tensorflow>=2.0 matplotlib numpy"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[31mERROR: albumentations 0.1.12 has requirement imgaug<0.2.7,>=0.2.5, but you'll have imgaug 0.2.9 which is incompatible.\u001b[0m\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1LEzzRPjrHzJ",
        "colab_type": "text"
      },
      "source": [
        "Import statemets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DRk8v2zUgLgY",
        "colab_type": "code",
        "outputId": "902fb797-ebc0-42cb-9b61-29c5f014364e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "#from keras.preprocessing import image # for data augmentation I think\n",
        "from keras.models import Sequential # For sequential models, not functional\n",
        "from keras.layers import Dense, Activation,MaxPooling2D, Dropout, Conv2D, Flatten # the layers used\n",
        "from keras.datasets import fashion_mnist\n",
        "from keras.utils import to_categorical # one-hot encoding\n",
        "\n",
        "import numpy as np\n",
        "import tensorflow as tf # tensorflow is the backend\n",
        "print( 'Tensorflow version: ' + tf.__version__) #+ \"  Keras version \" + print(keras.__version__) )"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Tensorflow version: 1.13.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XXk-QeSa7M25",
        "colab_type": "text"
      },
      "source": [
        "Set variables to enable / disable features depending on where the code is run"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lq37z-om7YhU",
        "colab_type": "code",
        "outputId": "1043fed3-c6be-42a1-a33d-2a16d70ab535",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "use_matplotlib, use_google_drive, use_gpu = True, True, True # False, False, True #\n",
        "\n",
        "if use_matplotlib:\n",
        "  import matplotlib.pyplot as plt\n",
        "\n",
        "# use gpu\n",
        "if use_gpu:\n",
        "  device_name = tf.test.gpu_device_name()\n",
        "  if device_name != '/device:GPU:0':\n",
        "    raise SystemError('GPU device not found')\n",
        "  print('Using GPU at: {}'.format(device_name))"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using GPU at: /device:GPU:0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tx_0GeqAsGXu",
        "colab_type": "text"
      },
      "source": [
        "# Load data & do some preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AZ4Y7HLoixEV",
        "colab_type": "code",
        "outputId": "a9aec5c0-8763-4b19-d574-1a9a3411f6e0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 208
        }
      },
      "source": [
        "# Load the dataset into training and test.\n",
        "(x_train, y_train), (x_test, y_test) = fashion_mnist.load_data()\n",
        "\n",
        "# Print training set shape - note there are 60,000 training data of image size of 28x28, 60,000 train labels)\n",
        "print(\"x_train shape:\", x_train.shape, \"y_train shape:\", y_train.shape)\n",
        "\n",
        "# Print the number of training and test datasets\n",
        "print(x_train.shape[0], 'train set')\n",
        "print(x_test.shape[0], 'test set')\n",
        "\n",
        "# Define the text labels\n",
        "fashion_mnist_labels = [\"T-shirt/top\",  # index 0\n",
        "                        \"Trouser\",      # index 1\n",
        "                        \"Pullover\",     # index 2 \n",
        "                        \"Dress\",        # index 3 \n",
        "                        \"Coat\",         # index 4\n",
        "                        \"Sandal\",       # index 5\n",
        "                        \"Shirt\",        # index 6 \n",
        "                        \"Sneaker\",      # index 7 \n",
        "                        \"Bag\",          # index 8 \n",
        "                        \"Ankle boot\"]   # index 9\n"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz\n",
            "32768/29515 [=================================] - 0s 3us/step\n",
            "Downloading data from http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz\n",
            "26427392/26421880 [==============================] - 2s 0us/step\n",
            "Downloading data from http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz\n",
            "8192/5148 [===============================================] - 0s 0us/step\n",
            "Downloading data from http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz\n",
            "4423680/4422102 [==============================] - 1s 0us/step\n",
            "x_train shape: (60000, 28, 28) y_train shape: (60000,)\n",
            "60000 train set\n",
            "10000 test set\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FFpw7yorqsWJ",
        "colab_type": "text"
      },
      "source": [
        "Normalization  & add a dimension for channel. Even if it is black & white images Keras wants a number for channels."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XLXfDHz0h2Oj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train = x_train.astype('float32') / 255\n",
        "x_test = x_test.astype('float32') / 255\n",
        "\n",
        "# the preprocess_input wants samples, x-values, y-values and channels\n",
        "# since black & white only one channel, so add a 1\n",
        "x_train = x_train.reshape(x_train.shape[0], 28, 28, 1)\n",
        "x_test = x_test.reshape(x_test.shape[0], 28, 28, 1)\n",
        "\n",
        "# \"One-hot-encode\" the labels in train and test\n",
        "y_train = to_categorical(y_train, 10)\n",
        "y_test = to_categorical(y_test, 10)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y0P_eWjrCvTp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Further break training data into training and validation sets, 5000 in validation set and remaining 55,000 for training set)\n",
        "(x_train, x_valid) = x_train[5000:], x_train[:5000] \n",
        "(y_train, y_valid) = y_train[5000:], y_train[:5000]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3cUbVvfWIwDW",
        "colab_type": "text"
      },
      "source": [
        "# Explore the data.\n",
        "Take a look at the y_train which, is now one-hot-encoded.\n",
        "Print the matrix of pixel values representing an image, then print the same data as an image."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mkLwP3amItip",
        "colab_type": "code",
        "outputId": "3089da53-a7ff-4d54-c4ee-a66a5ae11e63",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 807
        }
      },
      "source": [
        "# Choose one of the 60000 training samples\n",
        "img_index = 2\n",
        "# y_train contains the lables, ranging from 0 to 9\n",
        "print(\"Row number \" + str(img_index) +\" of the labels: \" + str(y_train[img_index]))\n",
        "label_index=np.argmax(y_train[img_index])\n",
        "# Print the label, for example 2 Pullover\n",
        "print (\"Category \"+str(label_index+1) + \" is \" +(fashion_mnist_labels[label_index]))\n",
        "\n",
        "# Show one of the images from the training dataset\n",
        "# reshaping neccessary to only show the columns with actual pixel data (not the one added with reshape above)\n",
        "img=x_train[img_index].reshape(28,28)\n",
        "np.set_printoptions(linewidth=125)\n",
        "print(\"The matrix of pixel values for sample \" + str(img_index) + \":\\n\" + str(np.around(img,1)))\n",
        "\n",
        "# show the same data but as an image\n",
        "if use_matplotlib:\n",
        "  plt.imshow(img)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Row number 2 of the labels: [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            "Category 8 is Sneaker\n",
            "The matrix of pixel values for sample 2:\n",
            "[[0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.  0.5 0.4 0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.5 0.5 0.5 0.6 0.8 0.7 0.6 0.3 0.1 0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.6 1.  0.6 0.6 0.9 0.7 0.7 0.6 0.7 0.5 0.2 0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.4 0.6 0.4 0.9 0.8 0.7 0.6 0.5 0.5 0.6 0.7 0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.3 0.6 0.9 0.3 0.5 0.8 0.8 0.8 0.4 0.4 0.4 0.5 0.5 0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.1 0.6 0.8 0.8 0.7 0.8 0.6 0.6 0.6 0.5 0.5 0.5 0.6 0.1]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.  0.1 0.  0.4 0.7 0.4 0.2 0.6 0.8 0.8 0.6 0.4 0.5 0.5 0.5 0.5 0.9 0.2]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.1 0.3 0.5 0.  0.4 0.5 0.6 0.9 0.6 0.6 0.6 0.4 0.3 0.3 0.4 0.4 0.7 0.1]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.1 0.2 0.4 0.  0.7 0.4 0.3 0.5 0.8 0.7 0.6 0.5 0.6 0.7 0.6 0.5 0.4 0.4 0.6 0.2]\n",
            " [0.  0.  0.1 0.2 0.3 0.3 0.3 0.4 0.1 0.5 0.9 0.1 0.2 0.5 0.8 0.9 0.7 0.5 0.5 0.5 0.5 0.6 0.6 0.5 0.6 0.5 0.6 0.3]\n",
            " [0.  0.1 0.3 0.3 0.2 0.1 0.1 0.3 0.3 0.3 0.3 0.5 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.5 0.4 0.5 0.5 0.5 0.6 0.5 0.6 0.3]\n",
            " [0.  0.1 0.  0.1 0.1 0.1 0.4 0.4 0.3 0.5 0.5 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.5 0.5 0.5 0.5 0.7 0.6 0.6 0.7 0.4]\n",
            " [0.  0.1 0.1 0.1 0.1 0.1 0.2 0.3 0.4 0.5 0.5 0.5 0.5 0.6 0.7 0.7 0.6 0.6 0.6 0.5 0.4 0.4 0.6 0.7 0.6 0.5 0.5 0.3]\n",
            " [0.3 0.2 0.  0.1 0.2 0.1 0.1 0.1 0.  0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.2 0.1 0.1 0.1 0.1 0.1]\n",
            " [0.1 0.6 0.5 0.4 0.2 0.1 0.1 0.1 0.2 0.2 0.1 0.2 0.1 0.1 0.2 0.2 0.1 0.1 0.2 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1]\n",
            " [0.  0.  0.1 0.4 0.7 0.7 0.6 0.6 0.3 0.3 0.3 0.2 0.2 0.3 0.2 0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.3 0.3 0.3 0.3 0.4 0.1]\n",
            " [0.  0.  0.  0.  0.  0.  0.2 0.4 0.4 0.5 0.6 0.5 0.5 0.6 0.1 0.2 0.2 0.2 0.1 0.1 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]\n",
            " [0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0.  0. ]]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKYafyCmGn8gphp/IKYafyKn/ByhBka2Exz3fAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yT0wWpewrZg4",
        "colab_type": "text"
      },
      "source": [
        "# The model\n",
        "##Create the model and add layers"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nxUpYlskrcPM",
        "colab_type": "code",
        "outputId": "e4fc944d-df62-42ec-8eed-a5091b8f7448",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 593
        }
      },
      "source": [
        "my_model = Sequential()\n",
        "\n",
        "my_model.add( Conv2D(filters=64, kernel_size=(4,4), padding='same', activation='relu', input_shape=(28,28,1))) \n",
        "my_model.add( MaxPooling2D(pool_size=(2,2),strides=(2,2)))\n",
        "my_model.add( Dropout(rate=0.3))\n",
        "\n",
        "my_model.add( Conv2D(filters=64, kernel_size=(4,4), padding='same', activation='relu'))\n",
        "my_model.add( MaxPooling2D(pool_size=2))\n",
        "my_model.add( Dropout(0.3))\n",
        "\n",
        "my_model.add( Flatten())\n",
        "my_model.add( Dense(256, activation='relu'))\n",
        "#my_model.add( Dropout(0.5))\n",
        "my_model.add( Dense(10, activation='softmax'))\n",
        "\n",
        "# print the number of layers\n",
        "print('Number of layers: ' + str(len(my_model.layers)))\n",
        "\n",
        "# print sumary of the model and its layers\n",
        "my_model.summary()"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "Number of layers: 9\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 28, 28, 64)        1088      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 14, 14, 64)        0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 14, 14, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 14, 14, 64)        65600     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 7, 7, 64)          0         \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 7, 7, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 3136)              0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 256)               803072    \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 10)                2570      \n",
            "=================================================================\n",
            "Total params: 872,330\n",
            "Trainable params: 872,330\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vQUSC9dNLH7S",
        "colab_type": "text"
      },
      "source": [
        "##Compile the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vkGAQGRsKaEn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "my_model.compile(loss='categorical_crossentropy',\n",
        "             optimizer='adam',\n",
        "             #optimizer='SGD',\n",
        "             #optimizer='RMSProp',\n",
        "             metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s6bIaeBaPnD4",
        "colab_type": "text"
      },
      "source": [
        "## Training\n",
        "Save training to history"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vJhPxCzCS-X4",
        "colab_type": "code",
        "outputId": "24a09094-25c2-4880-a557-1471c4cc9b5c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 801
        }
      },
      "source": [
        "from keras.callbacks import ModelCheckpoint\n",
        "checkpointer = ModelCheckpoint(filepath='best.hdf5', verbose = 1, save_best_only=True)\n",
        "bs = 64\n",
        "ep = 10\n",
        "\n",
        "history = my_model.fit(x_train,\n",
        "         y_train,\n",
        "         batch_size=bs,\n",
        "         epochs=ep,\n",
        "         validation_data=(x_valid, y_valid),\n",
        "         callbacks=[checkpointer])"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 55000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "55000/55000 [==============================] - 9s 166us/step - loss: 0.4573 - acc: 0.8343 - val_loss: 0.2908 - val_acc: 0.8944\n",
            "\n",
            "Epoch 00001: val_loss improved from inf to 0.29079, saving model to best.hdf5\n",
            "Epoch 2/10\n",
            "55000/55000 [==============================] - 6s 100us/step - loss: 0.3044 - acc: 0.8890 - val_loss: 0.2673 - val_acc: 0.9008\n",
            "\n",
            "Epoch 00002: val_loss improved from 0.29079 to 0.26730, saving model to best.hdf5\n",
            "Epoch 3/10\n",
            "55000/55000 [==============================] - 5s 100us/step - loss: 0.2650 - acc: 0.9021 - val_loss: 0.2342 - val_acc: 0.9118\n",
            "\n",
            "Epoch 00003: val_loss improved from 0.26730 to 0.23417, saving model to best.hdf5\n",
            "Epoch 4/10\n",
            "55000/55000 [==============================] - 5s 99us/step - loss: 0.2412 - acc: 0.9103 - val_loss: 0.2360 - val_acc: 0.9118\n",
            "\n",
            "Epoch 00004: val_loss did not improve from 0.23417\n",
            "Epoch 5/10\n",
            "55000/55000 [==============================] - 6s 107us/step - loss: 0.2184 - acc: 0.9183 - val_loss: 0.2073 - val_acc: 0.9214\n",
            "\n",
            "Epoch 00005: val_loss improved from 0.23417 to 0.20732, saving model to best.hdf5\n",
            "Epoch 6/10\n",
            "55000/55000 [==============================] - 6s 108us/step - loss: 0.2024 - acc: 0.9243 - val_loss: 0.2085 - val_acc: 0.9222\n",
            "\n",
            "Epoch 00006: val_loss did not improve from 0.20732\n",
            "Epoch 7/10\n",
            "55000/55000 [==============================] - 5s 100us/step - loss: 0.1865 - acc: 0.9296 - val_loss: 0.1965 - val_acc: 0.9258\n",
            "\n",
            "Epoch 00007: val_loss improved from 0.20732 to 0.19647, saving model to best.hdf5\n",
            "Epoch 8/10\n",
            "55000/55000 [==============================] - 5s 99us/step - loss: 0.1740 - acc: 0.9337 - val_loss: 0.2145 - val_acc: 0.9198\n",
            "\n",
            "Epoch 00008: val_loss did not improve from 0.19647\n",
            "Epoch 9/10\n",
            "55000/55000 [==============================] - 5s 100us/step - loss: 0.1584 - acc: 0.9401 - val_loss: 0.2192 - val_acc: 0.9204\n",
            "\n",
            "Epoch 00009: val_loss did not improve from 0.19647\n",
            "Epoch 10/10\n",
            "55000/55000 [==============================] - 5s 99us/step - loss: 0.1523 - acc: 0.9422 - val_loss: 0.2126 - val_acc: 0.9212\n",
            "\n",
            "Epoch 00010: val_loss did not improve from 0.19647\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BRPq4Rn_ks42",
        "colab_type": "text"
      },
      "source": [
        "## Plot feature map"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TE95EpHciyRM",
        "colab_type": "text"
      },
      "source": [
        "mount google drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "efcr2EM6i5GY",
        "colab_type": "code",
        "outputId": "abec7879-a38d-4f35-deb9-1a5bf0a1a4db",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 124
        }
      },
      "source": [
        "if use_google_drive:\n",
        "  from google.colab import drive\n",
        "  drive.mount('/content/gdrive')"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vSzqfc9-TEAM",
        "colab_type": "text"
      },
      "source": [
        "## Plot success"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a48C1zmx7qcL",
        "colab_type": "text"
      },
      "source": [
        "Function to save image to google drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XTE7UQVJ7m3S",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import datetime\n",
        "def save_png(name,ep,bs):\n",
        "  date = datetime.datetime.now()\n",
        "  date = date.strftime(\"%y%m%d%H%M%S\")\n",
        "  print(date)\n",
        "  filename = name + '-' + date\n",
        "  if ep != 0:\n",
        "    filename += '-'+str(ep)+'epochs-'+str(bs)+'batch-size.png'\n",
        "  \n",
        "  plt.savefig('/content/gdrive/My Drive/'+filename, bbox_inches='tight')  \n",
        "  print('file probably saved to: ' + filename)\n",
        " # with open('/content/gdrive/My Drive/'+filename , 'w') as f:\n",
        " #   f.write('content')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k1e7z2BQPrDw",
        "colab_type": "code",
        "outputId": "5e36c62d-685f-4419-b604-397cdb87f4f5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 608
        }
      },
      "source": [
        "# print(type(history.history))\n",
        "# print(len(history.history))\n",
        "# print(len(history.history['loss']))\n",
        "# print(history.history)\n",
        "\n",
        "# Plot training & validation accuracy values\n",
        "plt.plot(history.history['acc'])\n",
        "plt.plot(history.history['val_acc'])\n",
        "plt.title('Model accuracy, epochs: ' + str(ep) + ', batch size: ' + str(bs))\n",
        "plt.ylabel('Accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "# plt.axis([0, len(history.history['loss']),0,1])\n",
        "plt.legend(['Train', 'Validaton'], loc='upper left')\n",
        "save_png('acc',ep,bs)\n",
        "plt.show()\n",
        "\n",
        "# Plot training & validation loss values\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('Model loss, epochs: '+ str(ep) + ', batch size: ' + str(bs))\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('Epoch')\n",
        "\n",
        "#plt.axis([0, len(history.history['loss']),0,1])\n",
        "plt.axis\n",
        "plt.legend(['Train', 'Validation'], loc='upper left')\n",
        "plt.show()\n"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "190601060427\n",
            "file probably saved to: acc-190601060427-10epochs-64batch-size.png\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KFmm13sO--ot",
        "colab_type": "text"
      },
      "source": [
        "**Visualize a feature map / activation**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nNlXxMlJlBP3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras import models\n",
        "\n",
        "# Extracts the outputs of the top 8 layers:\n",
        "layer_outputs = [layer.output for layer in my_model.layers[:6]]\n",
        "# Creates a model that will return these outputs, given the model input:\n",
        "activation_model = models.Model(inputs=my_model.input, outputs=layer_outputs)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g96HxbbsrvYU",
        "colab_type": "code",
        "outputId": "180613f0-b548-4d4f-8cf8-9bb9b1b69312",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 327
        }
      },
      "source": [
        "img_tensor = x_train[img_index]\n",
        "# reshape from 3d to 4d\n",
        "img_tensor = img_tensor.reshape(1,28,28,1)\n",
        "activations = activation_model.predict(img_tensor)\n",
        "first_layer_activation = activations[0]\n",
        "print(\"Dimensions in the first activation layer: \" + str(first_layer_activation.shape[3]))\n",
        "\n",
        "if use_matplotlib:\n",
        "  plt.matshow(first_layer_activation[0, :, :, 33], cmap='magma')\n",
        "  save_png('feature-map',0,0)\n",
        "  plt.show()"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Dimensions in the first activation layer: 64\n",
            "190601060428\n",
            "file probably saved to: feature-map-190601060428\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAECCAYAAAAYUakXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEQIBQEIgAEgIBAAJgQAg+V87L6QDSshFHAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 288x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oMVq_um-uD_x",
        "colab_type": "text"
      },
      "source": [
        "##Evaluate the model\n",
        "Load the model with the optimal number of epochs. Predict and evaluate on the test-set."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qGRkbfAbuBG5",
        "colab_type": "code",
        "outputId": "9ad2e451-0c63-4e34-c40c-59037c1b9a86",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 191
        }
      },
      "source": [
        "my_model.load_weights('best.hdf5')\n",
        "\n",
        "predictions = my_model.predict(x_test)\n",
        "for prediction in predictions[:3]:\n",
        "  print(prediction)\n",
        "  \n",
        "\n",
        "score = my_model.evaluate(x_test, y_test, verbose=1)\n",
        "print(my_model.metrics_names)\n",
        "print(score)\n",
        "# Show test accuracy\n",
        "print('Loss: ', score[0],', Accuracy:', score[1])"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[4.6097497e-08 1.9634514e-08 1.1863065e-09 1.1158708e-07 1.4092194e-10 8.1628577e-06 1.8058172e-08 9.0378002e-05\n",
            " 9.7342001e-10 9.9990129e-01]\n",
            "[8.6439162e-05 6.6998833e-09 9.9955648e-01 9.6410349e-07 2.7542039e-05 2.6115582e-10 3.2865774e-04 1.2749253e-09\n",
            " 2.9682178e-08 5.8719402e-10]\n",
            "[2.4681248e-09 9.9999988e-01 4.7854171e-10 8.5668603e-08 2.5175484e-09 2.5601156e-14 3.7037939e-09 1.2568721e-15\n",
            " 1.2045214e-12 6.9711252e-14]\n",
            "10000/10000 [==============================] - 1s 60us/step\n",
            "['loss', 'acc']\n",
            "[0.22027545638084411, 0.922]\n",
            "Loss:  0.22027545638084411 , Accuracy: 0.922\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hLJmyqsQi1fc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}